What’s wrong with this picture? We are falling in love with an adolescent AI
By Alan Jacobson, Systems Architect & Analyst
Wall Street is pricing AI upside. It is not pricing AI risk.
It is betting on AI…
- Before it proves it can tell the truth
- Before it proves it can admit uncertainty
- Before it proves it can pay for itself
- Before it proves it can be controlled
Reality check
- AI adoption stalled in 2025
Traffic and app downloads flattened. Users tried it and walked away.
Why users abandon AI
- Hallucinations: It doesn’t know how to say “I don’t know.” Instead, it is confidently wrong.
Even the best models hallucinate heavily
- This is not a solved problem. Gemini 3 hallucinates 88%
- Trust collapses fast when answers sound right but aren’t
We are anthropomorphizing a system that isn’t reliable
- Intimacy before accuracy is a category error
The physics don’t work
- Planned data centers require massive electricity and water
- Power and water for cooling are real constraints, not footnotes
The economics don’t work
- AI is high-touch, 1:1
- Advertising doesn’t scale in private conversations
- Do you want your AI selling to you while advising you?
Billing is fantasy
- Tokens are words
- Words ≠ compute
- You can’t bill compute by counting words
Everyone assumes a revenue model will appear
- Because social media found one
- But social media is one-to-many
- And AI is 1:1
- That changes everything
Apple will power Siri with Google’s Gemini in Spring 2026
- What happened to “Apple Intelligence?”
- Is Apple outsourcing intelligence the way Microsoft outsourced AI to OpenAI?
The structural error
We are falling in love with AI
- Before it is trustworthy
- Before it can pay for itself
- Before it is controllable
That’s what’s wrong with the picture.